Screening Using Artificial Neural Network of Additives for Cu-Zn Oxide Catalyst for Methanol Synthesis from Syngas

Abstract

The activity of Cu _ Zn oxide catalysts for methanol synthesis from syngas varies depending on the additives to the oxide, and optimum composition is sensitive to the reaction conditions. An artificial neural network (ANN) was applied to identify the most effective additives based on the experimental results already reported. The physicochemical characters of element X, such as ionic radii and ionization energy, and the activity of Cu _ Zn _ X oxide catalyst were correlated using the ANN. Twenty-two types of X were supplied for the training of the ANN, and 29 activities of Cu _ Zn _ X, the X of which was not included in the training data, were predicted. Beryllium was predicted as the most effective additive, which was verified experimentally.

Facile Tool for Prediction of Catalytic Activity — Cu-Lanthanoid Catalyst for Methanol Synthesis

“…As a result Sn was predicted and experimentally verified to suppress the methane formation. In the similar way, beryllium was predicted as the most effective additive of Cu/Zn for methanol synthesis from syngas, which was verified experimentally (Omata et al, 2005). In other case, La and Ce, Sc and Nd were predicted to promote the activity of Ni/α-Al 2 O 3 for oxidative reforming of methane.…”

Screening of Additives to a Co/SrCO3 Catalyst by Artificial Neural Network for Preferential Oxidation of CO in Excess H2

“…(i) methanol carbonylation with Ni-(Cu, Zr, V, Ag, Pb, Sb, Bi, Zn, Tl, K, Mg)/active carbon catalyst; 38 (ii) methanol synthesis with Cu-Zn-(Li, B, Mg, Al, Ca, Sc, Ti, Cr, Mn, Ga, Y, Zr, La, Ce, Pr, Nd, Sm, Gd, Dy, Ho, Er, Yb); 39 (iii) DME synthesis with Cu-Zn-(B, K, Nb, Re, Cd, Ce, Sm, Tl); 40 (iv) CO PROX reaction with Co-(B, K, Sc, Mn, Zn, Nb, Ag, Nd, Re, and Tl)/SrCO 3 ; 29 (v) oxidative reforming of methane with Ni-(Ca, B, Mn, Fe, P, Cd, Re, Ce, Gd)/R-Al 2 O 3 . 41 For good prediction, the training data should be dispersed in the periodic table.…”

Design and Development of Cu−Zn Oxide Catalyst for Direct Dimethyl Ether Synthesis Using an Artificial Neural Network and Physicochemical Properties of Elements

“…Selection of Elements and Physicochemical Properties for Training. In our previous studies, the number of training data for ANN was not closely related to the quality of its prediction. , In both cases, RBFN was used as ANN to correlate the physicochemical properties with the catalytic activity. The properties used in the former case were first ionization energy (IE, eV), heat of vaporization (HV, kJ/mol), melting point (MP, K), and atomic radius (AR, pm).…”

“…Recently, ANNs have been increasingly applied to catalyst development through the prediction of catalyst performance, such as activity, selectivity, and durability, from experimental results. Among them, a few successful cases have been reported where catalytic properties were predicted from physicochemical properties of the catalyst elements. , We also successfully predicted an effective additive both on Ni/active carbon catalyst for carbonylation of methanol and on Cu−Zn oxide catalyst for methanol synthesis . Our predictions were conducted based on the previous experimental results and physicochemical properties of related elements.…”

“…2,3 We also successfully predicted an effective additive both on Ni/active carbon catalyst for carbonylation of methanol 4 and on Cu-Zn oxide catalyst for methanol synthesis. 5 Our predictions were conducted based on the previous experimental results and physicochemical properties of related elements. Better selectivity was obtained in the case of Ni/active carbon catalyst using ANN trained by 11 pairs of training data.…”